In [23]:
review_rate = data['reviews.rating'].value_counts()
review_rate.plot.bar()
Out[23]:
<AxesSubplot:>
import numpy as np
import pandas as pd
import nltk
from nltk.corpus import stopwords
from nltk.classify import SklearnClassifier
from nltk import sent_tokenize, word_tokenize, pos_tag
from nltk.tokenize import RegexpTokenizer
from nltk.stem import WordNetLemmatizer, PorterStemmer
from nltk import FreqDist
nltk.download('stopwords')
nltk.download('punkt')
[nltk_data] Downloading package stopwords to [nltk_data] /Users/marcofuentes/nltk_data... [nltk_data] Package stopwords is already up-to-date! [nltk_data] Downloading package punkt to [nltk_data] /Users/marcofuentes/nltk_data... [nltk_data] Package punkt is already up-to-date!
True
from wordcloud import WordCloud, STOPWORDS
import matplotlib.pyplot as plt
from matplotlib.ticker import StrMethodFormatter
%matplotlib inline
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split
data = pd.read_csv('GrammarandProductReviews.csv')
data.info()
<class 'pandas.core.frame.DataFrame'> RangeIndex: 71044 entries, 0 to 71043 Data columns (total 25 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 id 71044 non-null object 1 brand 71044 non-null object 2 categories 71044 non-null object 3 dateAdded 71044 non-null object 4 dateUpdated 71044 non-null object 5 ean 39065 non-null object 6 keys 71044 non-null object 7 manufacturer 70903 non-null object 8 manufacturerNumber 70841 non-null object 9 name 71044 non-null object 10 reviews.date 70977 non-null object 11 reviews.dateAdded 71044 non-null object 12 reviews.dateSeen 71044 non-null object 13 reviews.didPurchase 32158 non-null object 14 reviews.doRecommend 60429 non-null object 15 reviews.id 32158 non-null float64 16 reviews.numHelpful 32508 non-null float64 17 reviews.rating 71044 non-null int64 18 reviews.sourceURLs 71044 non-null object 19 reviews.text 71008 non-null object 20 reviews.title 70568 non-null object 21 reviews.userCity 5410 non-null object 22 reviews.userProvince 449 non-null object 23 reviews.username 70948 non-null object 24 upc 71042 non-null object dtypes: float64(2), int64(1), object(22) memory usage: 13.6+ MB
data = data[['brand','categories','manufacturer', 'name', 'reviews.didPurchase', 'reviews.doRecommend', 'reviews.numHelpful', 'reviews.rating', 'reviews.text', 'reviews.title', 'reviews.username']]
data['reviews.text'].head(20)
0 i love this album. it's very good. more to the... 1 Good flavor. This review was collected as part... 2 Good flavor. 3 I read through the reviews on here before look... 4 My husband bought this gel for us. The gel cau... 5 My boyfriend and I bought this to spice things... 6 Bought this earlier today and was excited to c... 7 I bought this product for my husband and I to ... 8 My husband and I bought this for some extra fu... 9 Got as a surprise for my husband there is noth... 10 tried it with my husband and felt no different... 11 Purchased this thinking it would have some sor... 12 Bought this to enhance our time a bit, did abs... 13 I bought this after really liking the his n he... 14 First time purchase of this type of product. M... 15 I bought this and tried this with my wife but ... 16 I bought this product to spice things up with ... 17 I bought this to try to spice things up, but I... 18 I bought this because it had better reviews th... 19 It was my first time using k-y. It didn't do a... Name: reviews.text, dtype: object
data.head()
| brand | categories | manufacturer | name | reviews.didPurchase | reviews.doRecommend | reviews.numHelpful | reviews.rating | reviews.text | reviews.title | reviews.username | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Universal Music | Movies, Music & Books,Music,R&b,Movies & TV,Mo... | Universal Music Group / Cash Money | Pink Friday: Roman Reloaded Re-Up (w/dvd) | NaN | NaN | 0.0 | 5 | i love this album. it's very good. more to the... | Just Awesome | Joshua |
| 1 | Lundberg | Food,Packaged Foods,Snacks,Crackers,Snacks, Co... | Lundberg | Lundberg Organic Cinnamon Toast Rice Cakes | True | NaN | NaN | 5 | Good flavor. This review was collected as part... | Good | Dorothy W |
| 2 | Lundberg | Food,Packaged Foods,Snacks,Crackers,Snacks, Co... | Lundberg | Lundberg Organic Cinnamon Toast Rice Cakes | True | NaN | NaN | 5 | Good flavor. | Good | Dorothy W |
| 3 | K-Y | Personal Care,Medicine Cabinet,Lubricant/Sperm... | K-Y | K-Y Love Sensuality Pleasure Gel | False | False | NaN | 1 | I read through the reviews on here before look... | Disappointed | Rebecca |
| 4 | K-Y | Personal Care,Medicine Cabinet,Lubricant/Sperm... | K-Y | K-Y Love Sensuality Pleasure Gel | False | False | NaN | 1 | My husband bought this gel for us. The gel cau... | Irritation | Walker557 |
data['reviews.text'].head(20)
0 i love this album. it's very good. more to the... 1 Good flavor. This review was collected as part... 2 Good flavor. 3 I read through the reviews on here before look... 4 My husband bought this gel for us. The gel cau... 5 My boyfriend and I bought this to spice things... 6 Bought this earlier today and was excited to c... 7 I bought this product for my husband and I to ... 8 My husband and I bought this for some extra fu... 9 Got as a surprise for my husband there is noth... 10 tried it with my husband and felt no different... 11 Purchased this thinking it would have some sor... 12 Bought this to enhance our time a bit, did abs... 13 I bought this after really liking the his n he... 14 First time purchase of this type of product. M... 15 I bought this and tried this with my wife but ... 16 I bought this product to spice things up with ... 17 I bought this to try to spice things up, but I... 18 I bought this because it had better reviews th... 19 It was my first time using k-y. It didn't do a... Name: reviews.text, dtype: object
data.info()
<class 'pandas.core.frame.DataFrame'> RangeIndex: 71044 entries, 0 to 71043 Data columns (total 11 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 brand 71044 non-null object 1 categories 71044 non-null object 2 manufacturer 70903 non-null object 3 name 71044 non-null object 4 reviews.didPurchase 32158 non-null object 5 reviews.doRecommend 60429 non-null object 6 reviews.numHelpful 32508 non-null float64 7 reviews.rating 71044 non-null int64 8 reviews.text 71008 non-null object 9 reviews.title 70568 non-null object 10 reviews.username 70948 non-null object dtypes: float64(1), int64(1), object(9) memory usage: 6.0+ MB
data['brand'] = data['brand'].str.lower()
data['categories'] = data['categories'].str.lower()
data['manufacturer'] = data['manufacturer'].str.lower()
data['name'] = data['name'].str.lower()
data['reviews.text'] = data['reviews.text'].str.lower()
data['reviews.title'] = data['reviews.title'].str.lower()
data['reviews.username'] = data['reviews.username'].str.lower()
data['brand'] = data['brand'].str.replace(r'http\S+', ' ')
data['categories'] = data['categories'].str.replace(r'http\S+', ' ')
data['manufacturer'] = data['manufacturer'].str.replace(r'http\S+', ' ')
data['name'] = data['name'].str.replace(r'http\S+', ' ')
data['reviews.text'] = data['reviews.text'].str.replace(r'http\S+', ' ')
data['reviews.title'] = data['reviews.title'].str.replace(r'http\S+', ' ')
data['reviews.username'] = data['reviews.username'].str.replace(r'http\S+', ' ')
<ipython-input-13-6d5f99569204>:1: FutureWarning: The default value of regex will change from True to False in a future version. data['brand'] = data['brand'].str.replace(r'http\S+', ' ') <ipython-input-13-6d5f99569204>:2: FutureWarning: The default value of regex will change from True to False in a future version. data['categories'] = data['categories'].str.replace(r'http\S+', ' ') <ipython-input-13-6d5f99569204>:3: FutureWarning: The default value of regex will change from True to False in a future version. data['manufacturer'] = data['manufacturer'].str.replace(r'http\S+', ' ') <ipython-input-13-6d5f99569204>:4: FutureWarning: The default value of regex will change from True to False in a future version. data['name'] = data['name'].str.replace(r'http\S+', ' ') <ipython-input-13-6d5f99569204>:5: FutureWarning: The default value of regex will change from True to False in a future version. data['reviews.text'] = data['reviews.text'].str.replace(r'http\S+', ' ') <ipython-input-13-6d5f99569204>:6: FutureWarning: The default value of regex will change from True to False in a future version. data['reviews.title'] = data['reviews.title'].str.replace(r'http\S+', ' ') <ipython-input-13-6d5f99569204>:7: FutureWarning: The default value of regex will change from True to False in a future version. data['reviews.username'] = data['reviews.username'].str.replace(r'http\S+', ' ')
data['brand'] = data['brand'].str.replace(r'[^A-Za-z\s]+', ' ')
data['categories'] = data['categories'].str.replace(r'[^A-Za-z]+', ' ')
data['manufacturer'] = data['manufacturer'].str.replace(r'[^A-Za-z]+', ' ')
data['name'] = data['name'].str.replace(r'[^A-Za-z\s]+', ' ')
data['reviews.text'] = data['reviews.text'].str.replace(r'[^A-Za-z]+', ' ')
data['reviews.title'] = data['reviews.title'].str.replace(r'[^A-Za-z]+', ' ')
data['reviews.username'] = data['reviews.username'].str.replace(r'[^A-Za-z]+', ' ')
<ipython-input-14-4a2c5c0edbed>:1: FutureWarning: The default value of regex will change from True to False in a future version. data['brand'] = data['brand'].str.replace(r'[^A-Za-z\s]+', ' ') <ipython-input-14-4a2c5c0edbed>:2: FutureWarning: The default value of regex will change from True to False in a future version. data['categories'] = data['categories'].str.replace(r'[^A-Za-z]+', ' ') <ipython-input-14-4a2c5c0edbed>:3: FutureWarning: The default value of regex will change from True to False in a future version. data['manufacturer'] = data['manufacturer'].str.replace(r'[^A-Za-z]+', ' ') <ipython-input-14-4a2c5c0edbed>:4: FutureWarning: The default value of regex will change from True to False in a future version. data['name'] = data['name'].str.replace(r'[^A-Za-z\s]+', ' ') <ipython-input-14-4a2c5c0edbed>:5: FutureWarning: The default value of regex will change from True to False in a future version. data['reviews.text'] = data['reviews.text'].str.replace(r'[^A-Za-z]+', ' ') <ipython-input-14-4a2c5c0edbed>:6: FutureWarning: The default value of regex will change from True to False in a future version. data['reviews.title'] = data['reviews.title'].str.replace(r'[^A-Za-z]+', ' ') <ipython-input-14-4a2c5c0edbed>:7: FutureWarning: The default value of regex will change from True to False in a future version. data['reviews.username'] = data['reviews.username'].str.replace(r'[^A-Za-z]+', ' ')
data.head()
| brand | categories | manufacturer | name | reviews.didPurchase | reviews.doRecommend | reviews.numHelpful | reviews.rating | reviews.text | reviews.title | reviews.username | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | universal music | movies music books music r b movies tv movie b... | universal music group cash money | pink friday roman reloaded re up w dvd | NaN | NaN | 0.0 | 5 | i love this album it s very good more to the h... | just awesome | joshua |
| 1 | lundberg | food packaged foods snacks crackers snacks coo... | lundberg | lundberg organic cinnamon toast rice cakes | True | NaN | NaN | 5 | good flavor this review was collected as part ... | good | dorothy w |
| 2 | lundberg | food packaged foods snacks crackers snacks coo... | lundberg | lundberg organic cinnamon toast rice cakes | True | NaN | NaN | 5 | good flavor | good | dorothy w |
| 3 | k y | personal care medicine cabinet lubricant sperm... | k y | k y love sensuality pleasure gel | False | False | NaN | 1 | i read through the reviews on here before look... | disappointed | rebecca |
| 4 | k y | personal care medicine cabinet lubricant sperm... | k y | k y love sensuality pleasure gel | False | False | NaN | 1 | my husband bought this gel for us the gel caus... | irritation | walker |
tokenizer = RegexpTokenizer(r'\w+')
data['brand'] = data['brand'].apply(lambda x: tokenizer.tokenize(x))
data['categories'] = data['categories'].apply(lambda x: tokenizer.tokenize(x))
data['manufacturer'] = data['manufacturer'].apply(lambda x: tokenizer.tokenize(str(x)))
data['name'] = data['name'].apply(lambda x: tokenizer.tokenize(x))
data['reviews.text'] = data['reviews.text'].apply(lambda x: tokenizer.tokenize(str(x)))
data['reviews.title'] = data['reviews.title'].apply(lambda x: tokenizer.tokenize(str(x)))
data['reviews.username'] = data['reviews.username'].apply(lambda x: tokenizer.tokenize(str(x)))
stopwords = stopwords.words('english')
data['reviews.text'] = data['reviews.text'].apply(lambda y: [w for w in y if w not in stopwords])
data['reviews.title'] = data['reviews.title'].apply(lambda y: [w for w in y if w not in stopwords])
data.head()
| brand | categories | manufacturer | name | reviews.didPurchase | reviews.doRecommend | reviews.numHelpful | reviews.rating | reviews.text | reviews.title | reviews.username | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | [universal, music] | [movies, music, books, music, r, b, movies, tv... | [universal, music, group, cash, money] | [pink, friday, roman, reloaded, re, up, w, dvd] | NaN | NaN | 0.0 | 5 | [love, album, good, hip, hop, side, current, p... | [awesome] | [joshua] |
| 1 | [lundberg] | [food, packaged, foods, snacks, crackers, snac... | [lundberg] | [lundberg, organic, cinnamon, toast, rice, cakes] | True | NaN | NaN | 5 | [good, flavor, review, collected, part, promot... | [good] | [dorothy, w] |
| 2 | [lundberg] | [food, packaged, foods, snacks, crackers, snac... | [lundberg] | [lundberg, organic, cinnamon, toast, rice, cakes] | True | NaN | NaN | 5 | [good, flavor] | [good] | [dorothy, w] |
| 3 | [k, y] | [personal, care, medicine, cabinet, lubricant,... | [k, y] | [k, y, love, sensuality, pleasure, gel] | False | False | NaN | 1 | [read, reviews, looking, buying, one, couples,... | [disappointed] | [rebecca] |
| 4 | [k, y] | [personal, care, medicine, cabinet, lubricant,... | [k, y] | [k, y, love, sensuality, pleasure, gel] | False | False | NaN | 1 | [husband, bought, gel, us, gel, caused, irrita... | [irritation] | [walker] |
stemmer = PorterStemmer()
def word_stemmer(data):
stem_text = " ".join([stemmer.stem(i) for i in data])
return stem_text
data['reviews.text'].apply(lambda x: word_stemmer(x))
0 love album good hip hop side current pop sound...
1 good flavor review collect part promot
2 good flavor
3 read review look buy one coupl lubric ultim di...
4 husband bought gel us gel caus irrit felt like...
...
71039 deal dri eye sensit product eye clinic suggest...
71040 great stuff help keep contact moist day
71041 right price
71042 pleas scarf arriv multipl reason print color b...
71043 excel album make holiday better give nice chri...
Name: reviews.text, Length: 71044, dtype: object
data['reviews.text'] = data['reviews.text'].apply(lambda x: word_stemmer(x))
df = data['reviews.text'].str.split(expand=True).stack().value_counts()
review_rate = data['reviews.rating'].value_counts()
review_rate.plot.bar()
<AxesSubplot:>
Grafica de frecuencia para los diferentes ratings de reviews
all_text = data['reviews.text'].str.split(' ')
text_data = [" ".join(text) for text in all_text]
final_text = " ".join(text_data)
wordcloud_text = WordCloud(
width = 3000,
height = 2000,
background_color = 'black',
stopwords = STOPWORDS).generate(final_text)
fig = plt.figure(
figsize = (40, 30),
facecolor = 'k',
edgecolor = 'k')
plt.imshow(wordcloud_text, interpolation = 'bilinear')
plt.axis('off')
plt.tight_layout(pad=0)
plt.show()
!pip3 install textblob
from textblob import TextBlob
Collecting textblob
Downloading textblob-0.15.3-py2.py3-none-any.whl (636 kB)
|████████████████████████████████| 636 kB 1.3 MB/s eta 0:00:01
Requirement already satisfied: nltk>=3.1 in /Users/marcofuentes/opt/anaconda3/lib/python3.8/site-packages (from textblob) (3.6.1)
Requirement already satisfied: joblib in /Users/marcofuentes/opt/anaconda3/lib/python3.8/site-packages (from nltk>=3.1->textblob) (1.0.1)
Requirement already satisfied: click in /Users/marcofuentes/opt/anaconda3/lib/python3.8/site-packages (from nltk>=3.1->textblob) (7.1.2)
Requirement already satisfied: tqdm in /Users/marcofuentes/opt/anaconda3/lib/python3.8/site-packages (from nltk>=3.1->textblob) (4.59.0)
Requirement already satisfied: regex in /Users/marcofuentes/opt/anaconda3/lib/python3.8/site-packages (from nltk>=3.1->textblob) (2021.4.4)
Installing collected packages: textblob
Successfully installed textblob-0.15.3
def get_sentiment(text):
analysis = TextBlob(text)
# set sentiment
if analysis.sentiment.polarity > 0:
return 'positive'
elif analysis.sentiment.polarity == 0:
return 'neutral'
else:
return 'negative'
sentiments = data['reviews.text'].apply(lambda x: get_sentiment(x))
freq_sentiments = sentiments.value_counts()
freq_sentiments.plot.bar()
<AxesSubplot:>
data['sentiment'] = sentiments
data[['reviews.text', 'sentiment']]
| reviews.text | sentiment | |
|---|---|---|
| 0 | love album good hip hop side current pop sound... | positive |
| 1 | good flavor review collect part promot | positive |
| 2 | good flavor | positive |
| 3 | read review look buy one coupl lubric ultim di... | positive |
| 4 | husband bought gel us gel caus irrit felt like... | neutral |
| ... | ... | ... |
| 71039 | deal dri eye sensit product eye clinic suggest... | negative |
| 71040 | great stuff help keep contact moist day | positive |
| 71041 | right price | positive |
| 71042 | pleas scarf arriv multipl reason print color b... | positive |
| 71043 | excel album make holiday better give nice chri... | positive |
71044 rows × 2 columns
most_positive = data[(data['reviews.rating'] == 5) & (data['sentiment'] == 'positive')]
freq_positive = most_positive['name'].value_counts()
cont = 1
for i in freq_positive.index:
if cont < 11:
print(cont,' '.join(i))
cont+=1
1 clorox disinfecting wipes value pack scented ct total 2 the secret life of pets k uhd blu ray digital 3 tide original liquid laundry detergent oz 4 the jungle book blu ray dvd digital 5 clorox disinfecting bathroom cleaner 6 godzilla d includes digital copy ultraviolet d d blu ray dvd 7 l or al paris elvive extraordinary clay rebalancing conditioner fl oz 8 l oreal paris revitalift triple power deep acting moisturizer 9 independence day resurgence k uhd blu ray digital 10 tide pods detergent botanical rain ct
most_negative = data[(data['reviews.rating'] == 1) & (data['sentiment'] == 'negative')]
freq_negative = most_negative['name'].value_counts()
cont = 1
for i in freq_negative.index:
if cont < 11:
print(cont,' '.join(i))
cont+=1
1 rubbermaid reveal spray mop 2 tide original liquid laundry detergent oz 3 nexxus exxtra gel style creation sculptor 4 maybelline eye studio master duo glossy liquid liner black lacquer 5 independence day resurgence k uhd blu ray digital 6 olay total effects daily moisturizer in anti aging oz 7 clear scalp hair therapy total care nourishing shampoo 8 maybelline eye studio color plush shadow quad gutsy green 9 aveeno baby continuous protection lotion sunscreen with broad spectrum spf oz 10 sausage party k uhd digital
data['name'] = data['name'].apply(lambda x: ' '.join(map(str, x)))
names = data.drop_duplicates(subset="name")
most_comments = names['reviews.username'].value_counts()
cont = 1
for i in most_comments.index:
if cont < 11:
print(cont,' '.join(i))
cont+=1
1 an anonymous customer 2 donna 3 nan 4 lisa 5 mike 6 erica 7 sandy 8 susan 9 paula 10 unhappy
most_positive_comments = data[(data['sentiment'] == 'positive')]
user_most = most_positive_comments['reviews.username'].value_counts()
cont = 1
for i in user_most.index:
if cont < 11:
if len(i) > 0:
print(cont,' '.join(i))
cont+=1
1 an anonymous customer 2 chris 3 mike 4 anonymous 5 nan 6 john 7 lisa 8 jess 9 dave 10 james
most_negative_comments = data[(data['sentiment'] == 'negative')]
user_most = most_negative_comments['reviews.username'].value_counts()
cont = 1
for i in user_most.index:
if cont < 11:
if len(i) > 0:
print(cont,' '.join(i))
cont+=1
1 an anonymous customer 2 byamazon customer 3 chris 4 anonymous 5 mike 6 carol 7 karen 8 mary 9 steve 10 melissa
mostBrand_positive = data[(data['reviews.rating'] == 5) & (data['sentiment'] == 'negative')]
freq_positive = mostBrand_positive['brand'].value_counts()
cont = 1
for i in freq_positive.index:
if cont < 11:
print(cont,' '.join(i))
cont+=1
1 clorox 2 sony pictures 3 tide 4 l oreal paris 5 quilted northern 6 olay 7 universal home video 8 lemi shine 9 fox 10 warner home video
mostBrand_negative = data[(data['reviews.rating'] == 1) & (data['sentiment'] == 'negative')]
freq_negative = mostBrand_negative['brand'].value_counts()
cont = 1
for i in freq_negative.index:
if cont < 11:
print(cont,' '.join(i))
cont+=1
1 rubbermaid 2 tide 3 nexxus 4 maybelline 5 fox 6 l oreal paris 7 olay 8 clear scalp hair therapy 9 sony pictures 10 aveeno
most_negative = data[(data['reviews.rating'] == 1) & (data['sentiment'] == 'negative')]
freq_negative = most_negative['name'].value_counts()
negative_data = data[data['sentiment']== 'negative']
negative_data.head()
| brand | categories | manufacturer | name | reviews.didPurchase | reviews.doRecommend | reviews.numHelpful | reviews.rating | reviews.text | reviews.title | reviews.username | sentiment | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 6 | [k, y] | [personal, care, medicine, cabinet, lubricant,... | [k, y] | k y love sensuality pleasure gel | False | False | NaN | 1 | bought earlier today excit check base product ... | [disappointing] | [raeanne] | negative |
| 22 | [k, y] | [personal, care, medicine, cabinet, lubricant,... | [k, y] | k y love sensuality pleasure gel | False | True | NaN | 4 | use time stay bit sticki bad better product | [job] | [jo] | negative |
| 41 | [johnson, s] | [personal, care, baby, care, baby, bubble, bat... | [johnson, s] | johnson s baby bubble bath and wash oz | False | False | NaN | 1 | month old babi got extrem irrit product use le... | [bad, reaction] | [an, anonymous, customer] | negative |
| 43 | [johnson, s] | [personal, care, baby, care, baby, bubble, bat... | [johnson, s] | johnson s baby bubble bath and wash oz | False | True | 0.0 | 3 | would think conveni bubbl bath wash contain me... | [bubble, bath] | [gardenbunny] | negative |
| 63 | [olay] | [personal, care, skin, care, anti, aging, beau... | [p, g] | olay regenerist deep hydration regenerating cream | False | False | NaN | 1 | olay work overli dri skin found anyth work ski... | [product, smells, ok] | [piggyboy] | negative |
list_words = []
frequencies = {}
for index, item in negative_data.iterrows():
# print(index)
# print(item['reviews.text'])
list_words.extend(item['reviews.text'].split(' '))
# break
for word in list_words:
if word in frequencies:
frequencies[word] += 1
else:
frequencies[word] = 1
print(len(frequencies))
5888
cont = 0
for i in frequencies:
print(i, frequencies[i])
cont += 1
if cont > 15:
break
bought 341 earlier 4 today 35 excit 81 check 20 base 212 product 1682 descript 13 expect 106 someth 136 like 1082 regular 45 ky 1 fan 103 left 137 littl 300
sort_orders = sorted(frequencies.items(), key=lambda x: x[1], reverse=True)
for item in sort_orders[:25]:
print(item, get_sentiment(item[0]))
('use', 2293) neutral
('product', 1682) neutral
('review', 1085) neutral
('like', 1082) neutral
('part', 1006) neutral
('collect', 966) neutral
('promot', 898) neutral
('movi', 864) neutral
('one', 823) neutral
('would', 655) neutral
('get', 655) neutral
('time', 652) neutral
('bad', 623) negative
('skin', 611) neutral
('work', 601) neutral
('hair', 595) neutral
('tri', 508) neutral
('look', 508) neutral
('make', 500) neutral
('year', 484) neutral
('realli', 466) neutral
('day', 453) neutral
('well', 422) neutral
('buy', 395) neutral
('go', 384) neutral
all_text = most_negative['reviews.text'].str.split(' ')
text_data = [" ".join(text) for text in all_text]
final_text = " ".join(text_data)
wordcloud_text = WordCloud(
width = 3000,
height = 2000,
background_color = 'black',
stopwords = STOPWORDS).generate(final_text)
fig = plt.figure(
figsize = (40, 30),
facecolor = 'k',
edgecolor = 'k')
plt.imshow(wordcloud_text, interpolation = 'bilinear')
plt.axis('off')
plt.tight_layout(pad=0)
plt.show()
Tomando en cuenta el análisis previo y la nube de palabras de arriba, creo que seria una buena idea que la empresa mejore el precio, presenteacion, limpieza y el olor de sus productos.
Analizando la frecuencia de las palabras de las malas reviews, podemos darnos cuenta que algunas de las mas frecuentes son disappoint, smell, wash, y look. Todo esto nos sugiere que los productos que les estan llegando a los usuarios no estan pasando por todas las pruebas de calidad necesarias para asegurar su buena presentacion.
Es importante notar que no hay palabras que hagan alusion a un mal producto, solo a una mala impresión o presentación del mismo. Esto es util porque tratar de implementar mejoras en la calidad del producto en sí seria una pérdida de recursos.